Can i has underscores? :)

> I checked, that in Pharo image i could simply use:
>
> foo_bar
>
> a_b := 5.
> FOO_BAR := 10.
>
> ^ FOO_BAR
>
> and it works!
>
> It would be cool to have similar in squeak, even if it may require to
> define something like:
>
> MyClass class>>canIHasUnderscores
>    ^ true
>
> It would be really nice to have, especially for things like OpenGL
> API, which using underscores in constant names.
> Exterminating underscores in those identifiers makes them really hard
> to read , and awkward to use, because
> every time you going to code something, first you usually refer to
> docs, where you see these names with underscores,
> but then once you going to type this name it costs you an additional
> mental effort - don't forget to skip underscores.
>

> I checked, that in Pharo image i could simply use:
>
> foo_bar
>
>        a_b := 5.
>        FOO_BAR := 10.
>
>        ^ FOO_BAR
>
> and it works!
>
> It would be cool to have similar in squeak, even if it may require to
> define something like:
>
> MyClass class>>canIHasUnderscores
>  ^ true
>
> It would be really nice to have, especially for things like OpenGL
> API, which using underscores in constant names.
> Exterminating underscores in those identifiers makes them really hard
> to read , and awkward to use, because
> every time you going to code something, first you usually refer to
> docs, where you see these names with underscores,
> but then once you going to type this name it costs you an additional
> mental effort - don't forget to skip underscores.
>
> --
> Best regards,
> Igor Stasenko AKA sig.
>
>

> Alex
>
> 2010/4/14 Igor Stasenko <[hidden email]>:
>> I checked, that in Pharo image i could simply use:
>>
>> foo_bar
>>
>>        a_b := 5.
>>        FOO_BAR := 10.
>>
>>        ^ FOO_BAR
>>
>> and it works!
>>
>> It would be cool to have similar in squeak, even if it may require to
>> define something like:
>>
>> MyClass class>>canIHasUnderscores
>>  ^ true
>>
>> It would be really nice to have, especially for things like OpenGL
>> API, which using underscores in constant names.
>> Exterminating underscores in those identifiers makes them really hard
>> to read , and awkward to use, because
>> every time you going to code something, first you usually refer to
>> docs, where you see these names with underscores,
>> but then once you going to type this name it costs you an additional
>> mental effort - don't forget to skip underscores.
>>
>> --
>> Best regards,
>> Igor Stasenko AKA sig.
>>
>>
>
>

> 2010/4/14 Igor Stasenko <[hidden email]>:
>> Is there an option to use a per-class setting, instead/in addition of
>> a system-wide one?
>
> Haven't tested this but Behavior>>allowUnderscoreSelectors suggests so:
>
>        "Return true if the receiver and its subclasses should be allowed to
> use underscore in selectors. Returning nil here means that the
> compiler should use the system-wide default preference. Also see
> #allowUnderscoreAssignments.
>
>        Combinations: If both preferences are false, underscore is illegal.
> If both preferences are true, underscore assignment needs the be
> surrounded by spaces and a single underscore cannot be used as
> selector."
>        ^nil
>

> On 14 April 2010 13:44, Alexander Lazarević <[hidden email]> wrote:
>> 2010/4/14 Igor Stasenko <[hidden email]>:
>>> Is there an option to use a per-class setting, instead/in addition of
>>> a system-wide one?
>>
>> Haven't tested this but Behavior>>allowUnderscoreSelectors suggests so:
>>
>>        "Return true if the receiver and its subclasses should be allowed to
>> use underscore in selectors. Returning nil here means that the
>> compiler should use the system-wide default preference. Also see
>> #allowUnderscoreAssignments.
>>
>>        Combinations: If both preferences are false, underscore is illegal.
>> If both preferences are true, underscore assignment needs the be
>> surrounded by spaces and a single underscore cannot be used as
>> selector."
>>        ^nil
>>
>
> Oh , well it seems works only for selectors, but not for ivars/classvars :(
> i put the
> MyClass class>>allowUnderscoreSelectors
>        ^ true
>
> on previously defined MyClass, and then tried to add ivar with underscores:
>
> Object subclass: #MyClass
>        instanceVariableNames: 'a_b_c'
>        classVariableNames: ''
>        poolDictionaries: ''
>        category: 'OpenGL-Specs'
>
> after accepting it, i got no error (what is strange), but instead ,
> the class having 3 ivars:
>
> Object subclass: #MyClass
>        instanceVariableNames: 'a b c'
>        classVariableNames: ''
>        poolDictionaries: ''
>        category: 'OpenGL-Specs'
>
>
> This is inconsistent. It should either throw an error or accept the
> variable name with underscore, but not silently turn
> 'a_b_c' into 3 separate ivars.
>
>> Alex
>>
>>
>
>
>
> --
> Best regards,
> Igor Stasenko AKA sig.
>
>

> Alexander Lazarević wrote:
>>
>> After setting the "Allow underscore selectors" preference to true in
>> 4.1.1 I just created a class MyClass with the method foo_bar and
>> MyClass new foo_bar returns 10?!
>>
>> I can't say why the note about the new sources file pops up on this
>> though.
>>
>> Alex
>>
>
> Hi Folks,
>
> I recently integrated Andreas' last changes to Cuis. Today I tried Igor's
> example and found a few problems, that the attached changes address. This
> stuff was done for Cuis, but should be easy to integrate in Squeak (and I
> guess Pharo too). Everybody feel free to integrate and test.
>
> 0489-ConfigurableUnderscoreShoutSupport-jmv.5.cs - Adds Shout support for
> #allowUnderscoreAssignments and #allowUnderscoreSelectos preference
> 0490-ConfigurableUnderscoreFixes-jmv.3.cs - Fixes some issues when parsing
> underscores in identifiers in .cs files. Does so by simplifying Scanner.
> Also makes #a_b to print correctly. With these fixes, such weird stuff as
> TestUnderscores.cs can be parser and installed.
> 0491-ConfigurableUnderscoreElsewhere-jmv.1.cs Fixes senders of #isLetter and
> #isAlphaNumeric that should also consider underscores.
>

> On 14 April 2010 20:10, Juan Vuletich <[hidden email]> wrote:
>  
>> Alexander Lazarević wrote:
>>    
>>> After setting the "Allow underscore selectors" preference to true in
>>> 4.1.1 I just created a class MyClass with the method foo_bar and
>>> MyClass new foo_bar returns 10?!
>>>
>>> I can't say why the note about the new sources file pops up on this
>>> though.
>>>
>>> Alex
>>>
>>>      
>> Hi Folks,
>>
>> I recently integrated Andreas' last changes to Cuis. Today I tried Igor's
>> example and found a few problems, that the attached changes address. This
>> stuff was done for Cuis, but should be easy to integrate in Squeak (and I
>> guess Pharo too). Everybody feel free to integrate and test.
>>
>> 0489-ConfigurableUnderscoreShoutSupport-jmv.5.cs - Adds Shout support for
>> #allowUnderscoreAssignments and #allowUnderscoreSelectos preference
>> 0490-ConfigurableUnderscoreFixes-jmv.3.cs - Fixes some issues when parsing
>> underscores in identifiers in .cs files. Does so by simplifying Scanner.
>> Also makes #a_b to print correctly. With these fixes, such weird stuff as
>> TestUnderscores.cs can be parser and installed.
>> 0491-ConfigurableUnderscoreElsewhere-jmv.1.cs Fixes senders of #isLetter and
>> #isAlphaNumeric that should also consider underscores.
>>
>>    
>
> Juan, i tried to file-in your changesets into a Squeak and got some errors:
>
> In  0490-ConfigurableUnderscoreFixes-jmv.3.cs
> - classDefinition: #Scanner category: #'Compiler-Kernel'
> is not a valid smalltalk expression
> next i tried to install things selectively and stumbled upon:
> Parser>>DNU: #classEncoding
> in debugger, somehow an encoder ivar holds an instance of Parser:
>
> allowUnderscoreAssignments
> "Query class + preference"
> ^encoder classEncoding allowUnderscoreAssignments
> ifNil:[super allowUnderscoreAssignments]
>
> my guess this is because of reshaping the Scanner class, which leads
> to a mess in existing Parser
> instances ivars.
>
> So, i think for a correct filein, it should create a separate
> NewScanner and NewParser classes,
> file them in, and only then , atomically rename them  to Scanner and Parser.
>
>  

> Igor Stasenko wrote:
>>
>> On 14 April 2010 20:10, Juan Vuletich <[hidden email]> wrote:
>>
>>>
>>> Alexander Lazarević wrote:
>>>
>>>>
>>>> After setting the "Allow underscore selectors" preference to true in
>>>> 4.1.1 I just created a class MyClass with the method foo_bar and
>>>> MyClass new foo_bar returns 10?!
>>>>
>>>> I can't say why the note about the new sources file pops up on this
>>>> though.
>>>>
>>>> Alex
>>>>
>>>>
>>>
>>> Hi Folks,
>>>
>>> I recently integrated Andreas' last changes to Cuis. Today I tried Igor's
>>> example and found a few problems, that the attached changes address. This
>>> stuff was done for Cuis, but should be easy to integrate in Squeak (and I
>>> guess Pharo too). Everybody feel free to integrate and test.
>>>
>>> 0489-ConfigurableUnderscoreShoutSupport-jmv.5.cs - Adds Shout support for
>>> #allowUnderscoreAssignments and #allowUnderscoreSelectos preference
>>> 0490-ConfigurableUnderscoreFixes-jmv.3.cs - Fixes some issues when
>>> parsing
>>> underscores in identifiers in .cs files. Does so by simplifying Scanner.
>>> Also makes #a_b to print correctly. With these fixes, such weird stuff as
>>> TestUnderscores.cs can be parser and installed.
>>> 0491-ConfigurableUnderscoreElsewhere-jmv.1.cs Fixes senders of #isLetter
>>> and
>>> #isAlphaNumeric that should also consider underscores.
>>>
>>>
>>
>> Juan, i tried to file-in your changesets into a Squeak and got some
>> errors:
>>
>> In  0490-ConfigurableUnderscoreFixes-jmv.3.cs
>> - classDefinition: #Scanner category: #'Compiler-Kernel'
>> is not a valid smalltalk expression
>> next i tried to install things selectively and stumbled upon:
>> Parser>>DNU: #classEncoding
>> in debugger, somehow an encoder ivar holds an instance of Parser:
>>
>> allowUnderscoreAssignments
>>        "Query class + preference"
>>        ^encoder classEncoding allowUnderscoreAssignments
>>                ifNil:[super allowUnderscoreAssignments]
>>
>> my guess this is because of reshaping the Scanner class, which leads
>> to a mess in existing Parser
>> instances ivars.
>>
>> So, i think for a correct filein, it should create a separate
>> NewScanner and NewParser classes,
>> file them in, and only then , atomically rename them  to Scanner and
>> Parser.
>>
>>
>
> Note that that file also includes definitions of #allowUnderscoreAssignments
> and #allowUnderscoreSelectors to deal with that situation. So, load them
> first.
>

> On 15 April 2010 15:09, Juan Vuletich <[hidden email]> wrote:
>  
>> Igor Stasenko wrote:
>>    
>>> On 14 April 2010 20:10, Juan Vuletich <[hidden email]> wrote:
>>>
>>>      
>>>> Alexander Lazarević wrote:
>>>>
>>>>        
>>>>> After setting the "Allow underscore selectors" preference to true in
>>>>> 4.1.1 I just created a class MyClass with the method foo_bar and
>>>>> MyClass new foo_bar returns 10?!
>>>>>
>>>>> I can't say why the note about the new sources file pops up on this
>>>>> though.
>>>>>
>>>>> Alex
>>>>>
>>>>>
>>>>>          
>>>> Hi Folks,
>>>>
>>>> I recently integrated Andreas' last changes to Cuis. Today I tried Igor's
>>>> example and found a few problems, that the attached changes address. This
>>>> stuff was done for Cuis, but should be easy to integrate in Squeak (and I
>>>> guess Pharo too). Everybody feel free to integrate and test.
>>>>
>>>> 0489-ConfigurableUnderscoreShoutSupport-jmv.5.cs - Adds Shout support for
>>>> #allowUnderscoreAssignments and #allowUnderscoreSelectos preference
>>>> 0490-ConfigurableUnderscoreFixes-jmv.3.cs - Fixes some issues when
>>>> parsing
>>>> underscores in identifiers in .cs files. Does so by simplifying Scanner.
>>>> Also makes #a_b to print correctly. With these fixes, such weird stuff as
>>>> TestUnderscores.cs can be parser and installed.
>>>> 0491-ConfigurableUnderscoreElsewhere-jmv.1.cs Fixes senders of #isLetter
>>>> and
>>>> #isAlphaNumeric that should also consider underscores.
>>>>
>>>>
>>>>        
>>> Juan, i tried to file-in your changesets into a Squeak and got some
>>> errors:
>>>
>>> In  0490-ConfigurableUnderscoreFixes-jmv.3.cs
>>> - classDefinition: #Scanner category: #'Compiler-Kernel'
>>> is not a valid smalltalk expression
>>> next i tried to install things selectively and stumbled upon:
>>> Parser>>DNU: #classEncoding
>>> in debugger, somehow an encoder ivar holds an instance of Parser:
>>>
>>> allowUnderscoreAssignments
>>>        "Query class + preference"
>>>        ^encoder classEncoding allowUnderscoreAssignments
>>>                ifNil:[super allowUnderscoreAssignments]
>>>
>>> my guess this is because of reshaping the Scanner class, which leads
>>> to a mess in existing Parser
>>> instances ivars.
>>>
>>> So, i think for a correct filein, it should create a separate
>>> NewScanner and NewParser classes,
>>> file them in, and only then , atomically rename them  to Scanner and
>>> Parser.
>>>
>>>
>>>      
>> Note that that file also includes definitions of #allowUnderscoreAssignments
>> and #allowUnderscoreSelectors to deal with that situation. So, load them
>> first.
>>
>>    
> 489 loaded without problems.. but then 490 fails.
>
>  
>> Cheers,
>> Juan Vuletich
>>
>>
>>